Supervised domain adaption for WSD
Eneko Agirre, Oier López de Lacalle · 2009
The lack of positive results on supervised domain adaptation for WSD have cast some doubts on the utility of hand-tagging general corpora and thus developing generic supervised WSD systems. In this paper we show for the first time that our WSD system trained on a general source corpus (Bnc) and the target corpus, obtains up to 22% error reduction when compared to a system trained on the target corpus alone. In addition, we show that as little as 40% of the target corpus (when supplemented with the source corpus) is sufficient to obtain the same results as training on the full target data. The key for success is the use of unlabeled data with svd, a combination of kernels and svm.